Shoresh Research Paper. An Overview of Israel s Education System and its Impact

Changing the paradigms Shoresh Research Paper An Overview of Israel s Education System and its Impact Abstract Israel is at a crossroads. It has one of the lowest productivity levels and the highest poverty rates in the developed world. With roughly half of its children receiving a Third World education, future economic sustainability is not a foregone conclusion. On the other hand, the country s leading universities are excellent, and they are converging with the top American universities. The knowledge needed to raise Israel to viable economic trajectories exists within its borders. But, an extremely inadequate education system is unable to channel this knowledge effectively to the primary and secondary schools, which in turn limits the ability to enter quality institutions of higher learning. This paper highlights aspects of education s economic impact and focuses on a number of misconceptions about the state of education in Israel. It provides an overview of achievements and failings in Israel s primary, secondary and higher education systems. December 2017 This policy research paper accompanies the Shoresh Handbook 2017-2018: Education and its Impact in Israel. It is intended to provide the larger context and a more thorough analysis of the issues covered in the Handbook. Prof. Dan Ben-David, President, Shoresh ; Department of Public Policy, Tel Aviv University; The author thanks Prof. Ayal Kimhi for his helpful comments. http://shoresh.institute

An Overview of Israel s Education System and its Impact 1. Introduction Israel s education system is, in many respects, an enigma. It is at once the home of some of the world s best universities, while also home to one of the worst primary and secondary education systems in developed countries. Anecdotal evidence suggests that Israel s education system or at least, large swathes of it were actually very good once. In lieu of actual data, this is not possible to confirm. 1 On paper, Israel has one of the most educated populations in the world, both in terms of average years of schooling and in terms of the share having an academic education. On the other hand, its productivity is not only low, it has been falling farther and farther behind the developed countries for decades, with the share of its population falling below the poverty line the highest in the developed world (Ben-David, 2016). Vast discrepancies in the educational level of parents across Israel s religious and ethnic divides are carried over into the next generation. As is the case in other countries, the level of pupil achievement is highly related to the education levels of the parents and, in particular, the mother s education. 1 It is possible that the economic situation of the country during its early decades did not offer much in the way of opportunities for many of the educated immigrants that reached its shores, causing a large number to choose teaching as their way of making ends meet. The free lunch experienced by Israel also extended to academia, where many of its first top scholars those who helped to set the subsequent high standards were born and trained abroad, at no cost to Israel, but to the great benefit of its future. http://shoresh.institute 1

The failure of Israel s education system in reducing gaps in comparison with other developed Figure 1 GAPS IN GRADES RELATIVE TO PUPILS WHOSE MOTHER MATRICULATED FROM HIGH SCHOOL PERCENT GAPS, BY MATERNAL EDUCATION LEVEL countries is highlighted in Figure 1, 15% which compares the PISA math achievements of pupils according to 3% 3% 2% 6% their mothers education. All comparisons are made to the test -6% -1% -4% -2% scores of children whose mothers matriculated from high school. In the ten countries with the highest PISA scores in 2015, children whose mothers did not have any formal education scored 6% below those -16% no education -20% primary -13% lower secondary Israel 10 leading countries with highest scores upper tertiary secondary non-academic (no matriculation) Ten leading countries: Canada, Estonia, Finland, Hong Kong, Japan, Korea, Netherlands, Singapore, Switzerland, Taiwan Source: Noam Gruber (Shoresh Institution research paper, 2017) Data: PISA 2012 academic degree whose mothers matriculated from high school. At the other end of the education spectrum, children whose mothers had an academic degree attained scores that were 6% above those of children with mothers who matriculated from high school. The impact of maternal education on the scores of Israeli pupils was considerably greater than the maternal impact in the ten leading countries. Pupils with mothers who did not study beyond primary school received math grades that were 20% below the grades of pupils whose mothers matriculated from high school. When mothers completed no more than lower secondary school, the gap among their children fell to 13%. Pupils with academically educated mothers attained scores that were 15% higher than those with mothers who only matriculated. A country s education system is its primary mechanism for reducing the heavy weights created by parental lack of education, which severely constrain their children s prospects of upward economic mobility as adults. Findings by Chetty, Friedman, Hilger, Saez, Schanzenbach and Yagan (2011) suggest that differences in school quality perpetuate http://shoresh.institute 2

income inequality. Chetty, Friedman and Rockoff (2011) estimate that an elimination of qualitative differences among American schools would reduce the intergenerational correlation of income by about a third. To get a better understanding of how such an ostensibly educated population has such low productivity and such high rates of poverty and inequality, this paper examines Israel s education system and its socioeconomic impact. Section Two provides some background and a review of previous studies that have examined the education system in Israel and elsewhere. The section analyzes quantitative measures of education and their impact on the labor market in section three. The fourth section turns to an examination of the quality of education in Israel and it economic impact. Section Five highlights the economic impact of improving math education at both the upper and lower ends of the achievement spectrum. Section Six illustrates basic facts and questions raised on the state of Israel s education. Section Seven examines the country s higher education system, while Section Eight concludes. 2. Background The theory of the firm holds that inputs are translated into output by means of a production function. Assessing the relationship between output and inputs over time requires an accurate measurement of inputs and output, adequately accounting for the endogeneity of the inputs while assuming that the production function does not change over time (except for technological advancement). The challenge of correctly defining inputs and output makes this task particularly complex when the focus turns to education. Can one rely solely on the size of the education budget to represent input? Given that the vast majority of the education budget goes toward teacher salaries, the question is: do teacher salaries, in a public education system characterized by rigid labor relations, reflect the quality of the teaching force? Can teaching force quality be measured in other ways? Can one rely on the number of pupils per class (which is affected by both wage spending and physical infrastructure) to reflect the quality of teaching? And how can the education system s output be measured? Is it sufficient to rely http://shoresh.institute 3

on quantitative indices, such as high school graduation rates, matriculation rates, and the percentage of persons with academic degrees, or should one also look at the skills and level of knowledge of the system s graduates? Alternatively, should the education system s output be measured via macroeconomic indices, such as growth and equality? These and other questions have preoccupied economists for years. They have generated an extensive literature on the importance of human capital, starting with the early studies of Mincer (1958), Schultz (1961) and Becker (1962), who coined the term. These were followed by analyses showing the importance of human capital to economic growth (e.g. Lucas, 1988, Romer, 1990; Mankiw, Romer and Weil, 1992; Hall and Jones, 1999), and continuing with research by Barro (2001), Heckman (2007), and Hanushek (2016), who focused on the distinction between quantity and quality of human capital, and on effective methods for teaching skills. In Israel, Helpman (1999) found that that an increase in the average number of school years accounted for 29% of the growth in the Israeli business sector s total factor productivity between 1971 and 1990. Bregman and Marom (2005) assessed the contribution of human capital to output via the production function approach, focusing on a panel of economic industries in the years 1970-1999. They found that a single additional year of schooling increased GDP and productivity by seven to eight percent. They also found that academic schooling has a greater effect on output and productivity and recommended strengthening Israeli higher education system. Argov (2016) utilized the growth accounting procedure to find that increased schooling since the 1970s has contributed one-third to one half of Israel s per capita GDP growth. However, many studies have found that educational quality contributes no less, and possibly more, than does educational quantity (e.g. Barro, 2001; Hanushek and Woessmann, 2015). Studies on the importance of human capital to worker income at the individual level also increasingly emphasize educational quality over educational quantity (e.g.: Chetty, Friedman, Hilger, Saez, Schanzenbach and Yagan, 2011). Academic institution quality (Weisbrod and Karpoff, 1968) and the field of study chosen (Leimieux, 2014; Kirkeboen, http://shoresh.institute 4

Leuven and Mogstad, 2016; Naylor, Smith and Telhaj, 2016; Rodriguez, Urzua and Reyes, 2016), subjects studied in high school (Altonji, 1995; Levine and Zimmerman, 1995; Rose and Betts, 2004; Morin, 2013; French et al., 2015), and preschool enrichment/early intervention programs (Crocker, Thomas, and Currie, 2002; Heckman, 2006) were all found to have an impact on labor market achievements. Several studies have been carried out in Israel on the relationship between educational quality and labor market achievements. Shwed and Shavit (2006) found occupational and income differences between graduates from Israel s research universities and non-research colleges, even after controlling for field of study. Zussman, Furman, Kaplan and Romanov (2009) found that the salaries of Israeli university graduates are higher than those of Israeli college graduates, for most study disciplines. Krill, Geva and Aloni (2016) discovered that the choice of academic field of study explains one-fifth of wage variability. Zussman and Tsur (2010) showed that graduates of high school vocational tracks have lower labor market achievements than do graduates of academic tracks. Kimhi and Horovitz (2015) found that upgrading high school math study from four to five units may increase future wages by about eight percent. The importance of educational quality raises the question of how countries can upgrade this area. While some may believe that larger budgets improve pupil achievements in public education systems, Hanushek and Kimko (2000) and Hanushek and Woessman (2008) found no such relationship between international exam achievements and the resources available to the relevant education systems during the years preceding the tests. Ben-David (2003, 2011) also demonstrated a similar lack of such a relationship with regard to Israel, finding relatively low levels of pupil achievements in comparison with other developed countries despite relatively high education expenditures per pupil. Lavy (2015a) found a positive relationship between instructional hours and pupil achievements on the PISA exams at the country level. Moreover, this relationship was stronger in countries that delegated greater degrees of administrative responsibility to the schools. In Israel, however, Ben-David (2012) showed that despite the fact that Israeli pupils http://shoresh.institute 5

receive more annual instructional hours than pupils in the majority of OECD countries, their achievements in international exams are below most of these same OECD countries. Most of the growth in Israel s education budget since 2007 has been channeled to teacher salary increases within the framework of new comprehensive wage bargaining agreements (Taub Center, 2015), but pupil performance on the international exams has improved minimally, if at all (Ben-David, 2015). Shavit and Feniger (2011) found that gaps in per pupil spending do not explain achievement gaps between Israeli pupils and their peers in other countries; rather, they link Israel s poor achievements to its larger class sizes. By contrast, Victor Lavy (1998) showed that funding gaps between Jewish and Arab schools were responsible for a major portion of the disparity in achievements in primary school math between Jewish and Arab pupils in 1991-1992. Budgetary gaps and socioeconomic differences explain 60% of math achievement gaps and exam failure rates during those years. Specifically, per pupil budgets, the number of instructional hours per pupil, and the share of certified teachers had a substantial impact on pupil achievements. In another study, Lavy (2012) used a 2004 change in school funding rules to quantify the impact of the per pupil budget on pupil achievements. He showed that the budget increases led to an extended school week and more instructional time spent on the core subjects of mathematics, science, and the English and Hebrew languages. This teaching upgrade resulted in improved pupil achievements in the relevant subjects. It was also found that the additional instructional hours increased the amount of time that pupils spent on homework, without reducing their satisfaction levels or negatively affecting their behavior. Angrist and Lavy (2002) found that introducing computers into primary and lower secondary classrooms during the 1990s did not improve pupil achievements in math. Teacher quality is a very important factor in pupil achievement levels (Navon and Shavit, 2012). Chetty, Friedman, Hilger, Saez, Schanzenbach and Yagan (2011) found that pupils who studied with more experienced preschool teachers reached higher income levels at age 27. Angrist and Lavy (2001) showed that a training program for primary school teachers http://shoresh.institute 6

in Jerusalem, in language and math skills, led to improved pupil achievements in secular schools. Financial incentives for teachers might be assumed to improve instructional quality, but research in this sphere has produced no unequivocal conclusions. Gamoran (2012) reviewed a number of American studies and concluded that financial incentives for teachers do not lead to better pupil performance. By contrast, a series of studies by Lavy attributed improved Israeli pupil achievements to teacher incentives. A program providing financial rewards to all teachers in schools with substantially improved pupil achievements led to a rise in the number of units studied for matriculation, higher matriculation exam scores, and higher matriculation eligibility rates, as well as reduced drop-out rates (Lavy, 2002). Another program, rewarding English and math teachers whose pupils matriculation exam achievements improved substantially, was found to have produced higher matriculation exam-taking rates, higher pass rates, and higher exam scores (Lavy, 2009). It was also found that this change was facilitated by different teaching methods, additional tutoring hours, and closer attention to pupils individual needs. A follow-up study (Lavy, 2015b) showed that the incentives impact persisted over the long term, as reflected in higher rates of academic study, higher employment, and higher wages. Other studies have looked at financial incentives for school principals and for the pupils themselves. Lavy (2008) found that wage increases for principals led to improved matriculation exam achievements. Angrist and Lavy (2009) showed that financial incentives for pupils led to improved matriculation achievements for girls, but not for boys. Some of the improvement was due to the allocation of more study time to matriculation exam preparation. Abramitzky and Lavy (2014) found that structural changes in the kibbutzim, which strengthened the connection between human capital and income, created incentives for kibbutz members to invest more in schooling. A discussion about improved instructional quality needs to also include the issue of class size. Smaller class sizes enable teachers to devote more time to each pupil and could lead to an improvement in discipline two factors that also reinforce one another. Studies on http://shoresh.institute 7

the impact of class size in other countries have not been uniform in their conclusions. For example, Chetty, Friedman, Hilger, Saez, Schanzenbach and Yagan (2011) found that pupils who attended smaller kindergarten classes were more likely to go on to higher education, but that their incomes at age 27 were not significantly higher. A recent study conducted in Norway found no significant relationship between class size and the income of pupils throughout their careers (Falch, Sandsør and Strøm, forthcoming). By contrast, Angrist and Lavy (1999) used an instrumental variable for class size in Israel and found that smaller class sizes substantially improved fifth grade reading and math scores. Achievements improved more moderately in fourth grade, while no improvement was found in third grade. Several studies have looked at the impact of school quality on pupil achievements. In one such study, Ellison and Swanson (2016) found substantial differences in the achievements of pupils at different schools, even when controlling for school and pupil characteristics. Woessmann (2016) showed that differences in education systems across countries are responsible for differences in pupil achievement. In Israel, Gould, Lavy and Paserman (2004) examined Ethiopian immigrants who had been randomly assigned to primary schools. They found that pupils who had been sent to schools with higher math achievements reached higher levels themselves, had a lower likelihood of dropping out of school and a higher probability of passing the matriculation exams. Another study by Gould, Lavy and Paserman, (2009) found that Israeli-born pupils in primary school classes with high shares of new immigrants reached lower achievements on matriculation exams. Shavit and Blank (2011) and Blank and Shavit (2013) examined the relationship between pupil achievements and class discipline levels, finding that as discipline improves, so do achievements. They also found that closing the discipline gap between Israel and a group of comparison countries would likely produce a substantial reduction in the achievement gap, though the disparity would not be entirely eliminated. Lavy and Schlosser (2011) showed that in classes with high percentages of girls, there were fewer disruptions, less violence, better teacher-pupil and pupil-pupil interactions, and less teacher burnout. All of these resulted in better scholastic performance. Lavy, Paserman and Schlosser (2011) http://shoresh.institute 8

found that a high percentage of low-ability pupils in a classroom has the opposite effect: poorer instructional quality, lower quality teacher-pupil and pupil-pupil interactions, more violence and class disruption all of which led to lower scholastic achievements. Lavy and Sand (2012) found that the presence of friends in class has a positive effect on test scores in English, math, and Hebrew language instruction, increases in the time devoted to homework, overall pupil satisfaction with the school, and a reduction in violent behavior. Lavy (2016) assessed the impact of teaching methods in English, Hebrew language, math and science on the grades of fifth graders (in 2002) and eighth graders (in 2005). He found that traditional methods are more effective in improving the scores of girls and pupils from socioeconomically disadvantaged backgrounds, while modern methods are similarly effective with all pupils. Interestingly, no relationship was found between pupil scores and other instructional characteristics, such as the teaching of individual study skills and teacherpupil relations. A key conclusion from this literature is that instructional quality and other features of the education system can be major factors in determining pupils future achievements which underscores the importance of having an effective educational policy. That said, policy cannot entirely ignore pupil background factors. For example, Gould, Lavy and Paserman (2011) followed pupils whose parents had immigrated to Israel from Yemen in the early years of Israeli statehood. They found that children who grew up in better living conditions were more likely to pursue higher education. Since there are substantial social and economic benefits of education at the national level that compound the private benefits to individuals, the policy challenge is to take the educational system in a direction that will minimize the impact of poor socioeconomic backgrounds on pupils achievements. http://shoresh.institute 9

3. Education quantity and Israel s labor market On the face of it, Israel s prime working age population is one of the most educated on the planet. 2 With 13.4 years of schooling per person between the ages of 35 and 54, the country is ranked third, after only the United States and Switzerland, each with an average of 13.5 years of schooling per person (Figure 2). To the extent that 12 years of schooling reflect completion of high school (though not necessarily graduation or matriculation), the majority of OECD countries are above this bar. Ten of the OECD countries sport a 13+ average, with an additional ten countries averaging at least 12 years of schooling. While gaps in average years of schooling among the majority of OECD countries are not particularly large, this is not the case when the focus turns to the share of the prime working age population with an academic degree. In six OECD countries, this share falls below 10% of the 35-54 year old population, with 11 additional countries averaging in the teens (Figure 3). In Israel, 31.6% of the prime working age population has an academic degree, placing it behind only three countries: the United States (32.6%), Ireland (33.9%), United States Switzerland Israel Czech Rep Slovakia Germany Ireland Canada Korea Denmark Estonia Japan United Kingdom Sweden Slovenia Hungary Australia Norway New Zealand Netherlands Poland Belgium Finland Greece France Luxembourg Latvia Spain Italy Iceland Austria Chile Mexico Portugal Turkey Figure 2 AVERAGE YEARS OF SCHOOLING PER PERSON 35-54 YEAR OLDS, 2010 6.7 8.8 8.5 13.5 13.5 13.4 13.4 13.3 13.2 13.2 13.2 13.0 13.0 12.9 12.9 12.7 12.5 12.4 12.3 12.2 12.1 12.1 12.1 11.9 11.7 11.7 11.6 11.6 11.6 11.2 11.2 11.0 10.9 10.5 10.4 Data: Barro and Lee (2016) and South Korea (37.6%). 2 While prime working age is generally considered between the ages of 25 and 54, this range is a bit more complicated in Israel. Because of compulsory military service, many Israelis delay their entry into higher education, thus pushing back graduation and full entry into the labor market until much later than is common in other countries. Consequently, the prime working age population considered here will be the 35-54 year old group. http://shoresh.institute 10

This shift in educational attainment over the years has been dramatic. Figure 4 displays the distribution of Israel s prime working age population by years of schooling. Haredim (ultra-orthodox Jews) are listed separately in Figure 4 and not divided by years of schooling because nearly all Haredi men do not study a core curriculum for more than 8 years and even Korea Ireland United States Israel Canada Greece Japan Switzerland Australia Estonia Belgium Luxembourg Denmark Netherlands Sweden Spain United Kingdom New Zealand Germany Norway Hungary Iceland Slovenia Finland Mexico Poland Austria France Latvia Italy Czech Rep Chile Slovakia Turkey Portugal 3.8% 9.4% 9.1% 8.8% 8.7% 6.9% Figure 3 SHARE OF PERSONS WITH A TERTIARY DEGREE 35-54 YEAR OLDS, 2010 25.9% 25.6% 25.2% 23.8% 23.7% 22.8% 21.5% 21.3% 20.7% 20.7% 20.2% 18.4% 18.3% 18.1% 17.5% 17.3% 17.0% 14.2% 13.4% 13.3% 12.9% 12.8% 33.9% 32.6% 31.6% 31.3% 31.1% 29.9% Data: Barro and Lee (2016) 37.6% then, it is just a partial one that excludes science, English and other basic material. After that, they continue with Torah studies for many years, and sometimes decades. Consequently, they are listed in the data as having 16+ years of education when this clearly designates something else entirely than for the remainder of the population. In 1970, 60% of Israel s prime working age non-haredi population had no more than 8 years of schooling. This share fell precipitously, to under 10%, by 2015. The share of persons with 16+ years of schooling has risen steadily over the decades, from having been the smallest in 1970 to its current position as the largest education group in Israel. 35% 30% 25% 20% 15% 10% 5% Figure 4 EDUCATION GROUPS, BY YEARS OF SCHOOLING AS SHARE OF TOTAL 35-54 YEAR OLD POPULATION, 1970-2015 9-10 haredim 5-8 0-4 0% 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 As of 2012, the Central Bureau of Statistics changed the estimation methodology in labor force surveys. Since 2012, Haredim are selfidentified in the data. Data by school years before 1979 includes Haredim. Source: Dan Ben-David and Oren Tirosh, Shoresh Institution Data: Israel s Central Bureau of Statistics 16+ 11-12 13-15 http://shoresh.institute 11

As Israel s economy has grown, it has shifted away from productive sectors necessitating little to no education, such as agriculture and textiles toward services and hi-tech sectors requiring higher levels of education and skills. The analogous shift in labor demand 100% 90% 80% 70% 60% 50% 40% 30% Figure 5 MALE EMPLOYMENT RATES, 1970-2015 BY EDUCATION LEVELS, 35-54 YEAR OLDS 16+ 11-12 9-10 5-8 Haredim (all years of educ) 0-4 over the years is vividly displayed in Figure 5. Although the share of prime working age men with no more than four years of schooling fell from 25% 20% 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 As of 2012, the Central Bureau of Statistics changed the estimation methodology in labor force surveys. Data by school years in 1970-1978 includes Haredim. Since 1979, it excludes Haredim. Source: Dan Ben-David and Oren Tirosh, Shoresh Institution Data: Israel s Central Bureau of Statistics to 2%, indicating a marked decline in their supply, demand for such men has almost evaporated. Their rates of employment have been in a near free-fall until recently. While there was almost no difference in 1970 s employment rates between the various education groups, a substantial gap has developed between them over the decades. Although the share of men with 16+ years of schooling has risen sharply during this span, demand for them has risen as well, with employment rates in 2015 over 90%, roughly where they were in 1970. For all other education groups, the lower the level of education, the greater the fall in employment rates a trend that can only be expected to continue, despite the aberration in recent years among the less educated as a result of significant cuts in welfare benefits. While the share of non-haredi men with very low levels of education has declined over the years, the share of Haredi men most of whom have no more than 8 years of formal education including a core curriculum has been rising. Two factors have contributed to the declining Haredi employment rates: (a) changing demand that has shifted away from the less skilled and poorly educated, and (b) greater welfare benefits that reduced supply. http://shoresh.institute 12

Though there is no reason to assume that Haredi education has become worse over the years, the fact that Israel s economy has evolved in a direction requiring better education and greater skills has translated into a declining employment share among Haredi men who are increasingly being left behind. Over 80% of them were employed in 1979. By the last decade, this share fell to less than 40%. While Haredim account for about 7% of the prime working age population, they account for 19% of all first-graders, indicating just how rapidly this group is growing and how quickly this issue may escalate into unmanageable proportions, if not addressed. The decline in Haredi employment was enhanced by their rising political power, which translated into substantial increases in welfare and other assistance that could enable so many men to choose life-styles of non-work. For example, average income supplements per recipient and average child benefits per recipient grew by 57% each in real terms between 1983 and 2001. Israel s major recession in the early 2000s led to extensive budget cuts, including many welfare benefits. By 2005, average income supplements per recipient had fallen by 26% while average child benefits per recipient fell by 46% (Ben-David, 2016). Over the next decade, average income supplements per recipient recovered a bit, rising by 6% through 2015, while average child benefits per recipient fell a further 13%. As Ben-David shows, these large reductions in benefits were accompanied by substantial changes in the labor market particularly among the less educated. Among the prime working age population with 16 or more years of education, each addition of 100 persons to this population was accompanied by an increase of 95 additional employed persons from 2002 to 2015. By contrast, in the prime working age population with less years of schooling, each increase of 100 persons in the population was accompanied by an increase of 270 employed persons. In other words, many less educated persons who had not worked before began to enter the labor market, including many Haredim, since the early 2000s. The importance of education for female employment is shown in Figure 6. Despite the increase in supply of more educated women, corresponding increases in demand have led http://shoresh.institute 13

to higher employment rates. There has always been a gap between employment rates of the various education groups, though this gap has widened over the decades. As Kimhi 100% 90% 80% 70% Figure 6 FEMALE EMPLOYMENT RATES, 1970-2015 BY EDUCATION LEVELS, 35-54 YEAR OLDS 16+ 11-12 (2012) shows, the primary source of the vastly increased female employment rates at the national level is due to more and more women moving up the education ladder into successively higher education groups that are characterized by higher employment rates. 60% 50% 40% 30% 20% 10% 9-10 0% 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 As of 2012, the Central Bureau of Statistics changed the estimation methodology in labor force surveys. Source: Dan Ben-David and Oren Tirosh, Shoresh Institution Data: Israel s Central Bureau of Statistics 5-8 0-4 Education s impact on employment crosses gender and religious lines. Figure 7 compares employment rates among poorly educated and highly educated prime working age Jewish (non-haredi) and Arab Israelis, men and women. Only 12% of Arab Israeli women who did not complete high school and have no more than 11 years of schooling were employed in 2015. About twothirds of similarly educated Jewish women and men of both religions are employed. At the other end of the 80% Figure 7 EMPLOYMENT RATES, 2015 35-54 YEAR OLDS 94% 91% 66% 64% 69% 95% education spectrum, nearly all Jewish (95%) and Arab Israeli (94%) prime working age men with academic degrees are employed, alongside 91% of similarly educated Jewish women. Employment rates among Arab Israeli 12% 0-11 Aca- years of demic school degree 0-11 Aca- years of demic school degree Arab Israeli women Jewish non-haredi women Arab Israeli men 0-11 Aca- years of demic school degree 0-11 Aca- years of demic school degree Jewish non-haredi men Source: Dan Ben-David and Oren Tirosh, Shoresh Institution Data: Israel s Central Bureau of Statistics http://shoresh.institute 14

women are lower (80%), but are, nonetheless, in a different league from what is transpiring among the less educated Arab Israeli women. The conventional wisdom in Israel, that Arab Israeli women do not work, is based primarily on the fact that a very large share of such women 100% 95% 90% 85% 80% 75% 70% 65% Figure 8 GAP IN HOURLY WAGES BETWEEN EMPLOYEES WITH 13+ AND 0-12 YEARS OF SCHOOLING, 1999-2015 AMONG FULL-TIME EMPLOYEES AGES 35-54 have low levels of education. Higher levels of education lead to substantially higher employment rates for all population groups. 60% 55% 50% 1999 2001 2003 2005 2007 2009 2011 2013 2015 Source: Ayal Kimhi, Shoresh Institution and Hebrew University Data: Israel s Central Bureau of Statistics Higher levels of education also lead to higher wages. Figure 8 displays the gap in hourly wages between full-time prime working age employees with 12 years of schooling at most and those with 13 or more years of schooling. From 1999 to 2007, the 13+ group earned approximately 60% more than the 0-12 group. Since then, this gap has risen substantially, reaching 95% by 2015. The spectrum of hourly wages for full-time, prime working age employees can be seen in Figure 9. Employees with Figure 9 HOURLY WAGES BY EDUCATION LEVEL FULL-TIME EMPLOYEES, AGES 35-54 BY HIGHEST ATTAINED LEVEL OF EDUCATION, 2015 130 no education, and those with a primary school education at most, receive 33 and 86 100 35 shekels per hour, respectively. As education levels rise, so do hourly wages, 33 35 45 52 58 reaching 86 shekels per hour for employees with BAs, 100 shekels per hour for those with MAs, and 130 shekels none primary secondary bagrut postsecondary high school matriculation BA MA PhD per hour for employees with PhDs. Source: Ayal Kimhi, Shoresh Institution and Hebrew University Data: Israel s Central Bureau of Statistics http://shoresh.institute 15

An examination of academic Figure 10 degree attainment rates and their changes over time (Figure 10) provides a glimpse of the high 60% 55% 50% SHARE OF ISRAELIS WITH ACADEMIC DEGREES, 2004-2015 AGES 35-54 secular Jews women men degree of disparity between key 45% population groups in Israel. Secular (non-religious) Jews comprise the 40% 35% 30% 25% religious Jews women men women most highly educated group in the country. Not only is the this group s share of individuals with academic degrees substantially higher than in all of the other groups, it has also been rising over the past decade for the women, and 20% 15% 10% 5% Arabic speaking Israelis Haredim 0% 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Non-religious Jews ultra-orthodox Jews Jews ranging from traditional through Orthodox Data: Israel s Central Bureau of Statistics men men women during most of the past decade with some stagnation in recent years for the men. Half of prime working age secular Jewish men in Israel have academic degrees, while nearly 60% of the secular Jewish women have attained one, as well. Religious Jews those spanning the spectrum between secular and Haredi Jews had considerably lower rates of academic attainment than secular Jews in the mid-2000s. Since then, the share of prime working age religious men and women has climbed steadily, by almost a half (from 23% to 34%) for the men and over two-fold (from 18% to 38%) for the women. As in the case of secular Jews, and as is prevalent in much of the developed world, the share of religious women with academic degrees is greater than the men s share. The prevailing wisdom in Israel with regard to Haredim is that the country has begun to turn the corner on their academic education, with a plethora of new higher education possibilities opening up to Haredim and substantial increases in their registration to these institutions. A more precise examination of the evidence suggests that this is no more than a misconception. http://shoresh.institute 16

While the sample is small, leading to high annual variability, the share of prime working age Haredim with an academic degree is both very low and has also been relatively constant over the past decade. There was an increase in 2015 that may, or may not, signal a change in trend for the future. Two commonly overlooked factors underlie the perception of a major improvement that is not borne out by the data. The first is that the Haredi population is growing very quickly, so there are many more Haredim in general, and not just those trying their luck in academia. This is why the figure focuses on population shares rather than on absolute numbers. Second, simply starting down the academic road does not imply successful completion of the entire route and the eventual attainment of an academic degree. The very poor primary and secondary school education that Haredim receive (men, in particular) acts a major stumbling block when they become adults and contributes to more Haredim dropping out than those completing an academic degree. The fact that Haredim in other countries are not allowed to deprive their children of a complete core curriculum as mandated by law in all of the developed world except in Israel provides Haredim in those countries with better prospects of successfully completing academic degrees. For example, 25% of American Haredim above the age of 20 have an academic degree (PEW, 2015), which is low compared to other population groups in the United States, but just over double the 12% rate for Israeli Haredim (Ben-David, 2015). The population group with the lowest rates of academic attainment are Arabic speaking Israelis, though their rates of academic degree attainment have been steadily climbing since the mid-2000s. This is especially true for the women, rising from 6% in 2004 to 15% in 2015 (from 13% to 16% for the men). There are vast differences within the Arabic speaking group. In 2004, 10% of prime working age Muslim men attained an academic degree, compared to 15% of the Druze men. Christian Arab Israelis, whose primary and secondary education system is considerably better than that of the Muslims and Druze, also have much higher academic completion rates, with 23% of these men receiving degrees in 2004. In all of these groups, the increases in http://shoresh.institute 17

academic degrees per capita were substantial: rising to 14% for Muslim men, 22% for Druze men, and 31% for Christian Arab men. The increases for women in two of the three Arab speaking groups was even greater than the increases exhibited by the men. Muslim women saw an almost three-fold increase within just over a decade in the share with academic degrees, from the extremely low 4.6% in 2004 to the still low but considerably higher 13.5% in 2015. The share of prime working age Druze women with academic degrees rose from 6.7% to 10.8% during the same time span. Christian Arab woman exhibited a doubling in their rates of academic attainment, from 16.4% in 2004 to 22.1% in 2015, becoming the only Arabic speaking community in Israel in which there is a greater share of women than men with academic degrees. Human capital plays a major role in determining the amount produced per hour of work, or labor productivity. In the final analysis, labor productivity determines how high hourly wages can rise. This positive link between hourly wages and hourly production across OECD countries (Figure 11) is visually apparent, despite the relative outliers (Luxembourg and Ireland). Clearly, other factors may play a role in loosening this relationship, but it does appear to hold in general. What transpired within Israel since 1999 is a case in point. The major recession that the country underwent in the early 2000s led to wage reductions in both the 0-12 and 13+ education groups. The wage paths of the two groups diverged in the recession s aftermath. While structural changes inherent in the growth process reduced demand for the less skilled and poorly educated, the major cuts in welfare benefits average hourly wages 35 30 25 20 15 10 5 0 Figure 11 RELATIONSHIP BETWEEN LABOR PRODUCTIVITY AND WAGES OECD COUNTRIES, CURRENT PPP-ADJUSTED DOLLARS, 2015 10 20 30 40 50 60 70 80 90 labor productivity (GDP per hour worked) Data: OECD http://shoresh.institute 18

discussed above led many of the less educated to enter the labor market for the first time thus driving up the supply of the less educated,further dampening their wages. Between 2006 and 2011, Figure 12 CUMULATIVE REAL CHANGE IN AVERAGE HOURLY WAGES AND IN LABOR PRODUCTIVITY FROM 1999 TO 2015 23% 24% this process resulted in a drop of over 10% in the wages of employees with 0-12 years of schooling. In the years since 2011, wages of this education group recovered, returning to a level just 1% below the 1999 levels. This outcome contrasts sharply with what has taken place among employees with 13+ years -1% 0-12 years of schooling 13+ years of schooling hourly wages of employees (by education level) Full-time employees, ages 35-54 labor productivity (GDP per hour worked) of schooling. The growth process led to increases in demand for this group after Israel emerged from Source: Ayal Kimhi, Shoresh Institution and Hebrew University Data: Israel s Central Bureau of Statistics its recession, resulting in substantial real wage increases. Overall, wages of employees with 13+ years of schooling rose by 23% from 1999 through 2015, almost identical to the 24% increase in labor productivity during this period (Figure 12). 4. Education quality and its economic impact Israel is uniquely placed among the family of nations to highlight why an emphasis on the quantity of education (as measured by the number of school years per person or the share of individuals with academic degrees) is insufficient. It has one of the most educated societies in the OECD when measured by education quantity, as opposed to quality alongside labor productivity that is below most OECD countries (Figure 13). Not only is its labor productivity relatively low, it has been steadily falling further and further behind the average labor productivity of the G7 countries since the 1970s (Figure 14) an outcome that is diametrically opposed to what one might expect in a country that caught up with, and even surpassed, nearly all of the developed world in terms of the quantitative education measures. http://shoresh.institute 19

Low physical capital investments also played a key role in dampening Israeli productivity (Ben-David 2017). Similarly, insufficient transportation infrastructure investments, causing extraordinarily heavy traffic congestion, further depressed productivity. For example, Israel has over three times the number of vehicles per kilometer road than the OECD average, though it has 38% fewer vehicles per capita (Ben- David 2015). Figure 13 LABOR PRODUCTIVITY IN 2015 GDP PER HOUR WORKED IN 35 OECD COUNTRIES, IN CURRENT PPP DOLLARS Luxembourg Ireland Norway Belgium Denmark United States France Netherlands Germany Switzerland Austria Sweden Finland Australia Italy United Kingdom Spain Canada Iceland Japan New Zealand Slovenia Israel Slovak Republic Czech Republic Turkey Portugal Greece Hungary Estonia Korea Poland Latvia Chile Mexico 72.1 69.7 68.3 67.6 67.1 66.6 65.6 61.8 60.5 56.5 56.0 53.6 52.4 51.3 50.8 45.7 43.0 42.0 41.5 41.1 40.8 39.1 38.6 36.0 34.9 34.6 33.0 31.9 31.2 29.1 26.1 20.1 82.3 Data: OECD 95.0 91.8 And yet, even within the education realm, an important ingredient is missing from the analysis thus far: education quality. Hanushek and Woessmann (2015) show that while there is a slight positive relationship between the average number of school years in a country and average annual growth rates (Figure 15), there is a much stronger positive relationship between the quality of a country s education (as measured by achievements in math, science and reading) and its 55 Figure 14 LABOR PRODUCTIVITY, 1970-2015 GDP PER HOUR WORKED IN 2010 PPP-ADJUSTED DOLLARS economic rate of growth (Figure 16). A year of schooling in Japan or 50 45 G7 17.2 Canada, for example, is not equivalent to 40 a year of education in Thailand or India or to a year of schooling in Israel. Gauging the quality of education in Israel is not as accurate is it could, or should be. On the one hand, the country requires all students wishing to graduate from high 35 30 25 20 5.4 Israel 15 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 Data: OECD, Israel s Central Bureau of Statistics, Bank of Israel http://shoresh.institute 20

school to pass a matriculation exam (bagrut, in Hebrew). These exams have been given for decades and could have been a very useful tool for determining whether education levels have improved or declined over the years. But the national exams have never been Adjusted average growth rate of GDP per capita 5% 4% 3% 2% Figure 15 YEARS OF SCHOOLING AND ECONOMIC GROWTH 50 COUNTRIES, 1960-2000 calibrated over time, rendering useless all intertemporal comparisons. To reduce the validity of these exams even further, final matriculation grades 1% 2 4 6 8 10 Adjusted years of schooling Relationship between both variables after controlling for the impact of other influencing variables (such as average years of schooling and initial level of real per capita GDP in 1960) on each. This yields an association between the two variables that is not driven by the control variables. The regression residuals are added to the unconditional means of the variables on each axis. Source: Eric Hanushek and Ludger Woessmann, (2015) include a local school component for which no attempt is made at calibration across schools. The result is a relatively expensive exam process that yields little to no usefulness in providing comparable benchmarks for measuring the quality of education provided in Israel. International exams, such Adjusted average growth rate of GDP per capita 6% 5% 4% 3% 2% 1% Figure 16 TEST SCORES AND ECONOMIC GROWTH 50 COUNTRIES, 1960-2000 as PISA, TIMSS and others used by Hanushek and Woessmann (2015) in their study, provide another route for gauging educational quality. But these 0 300 350 400 450 500 550 Adjusted test scores in core subjects Relationship between both variables after controlling for the impact of other influencing variables (such as average years of schooling and initial level of real per capita GDP in 1960) on each. This yields an association between the two variables that is not driven by the control variables. The regression residuals are added to the unconditional means of the variables on each axis. Source: Eric Hanushek and Ludger Woessmann, (2015) http://shoresh.institute 21

exams are not without their own problems especially when it comes to comparisons of Israel to other countries. Haredi boys, for example, do not study the material covered in these exams and do not participate in the exams. In general, the international exams stipulate that a country may exclude up to 5% of its pupils from the sample. This is usually intended for special needs pupils, or those living in distant rural areas. As shown in Figure 17, nearly all countries participating in the most recent TIMSS exam, administered in 2015, abided by these rules. A select few excluded slightly more than the acceptable exclusion rate. And then there is Israel, which excluded a full 23% of its pupils from the sample. Thus, in all of the international comparisons that follow, it can be assumed that the Israeli results reflect better outcomes than would have actually existed had all of the country s pupils participated in the exams. The international exams are calibrated on a per exam basis to a mean of 500. This mean remains 500 regardless of the year or the number of countries participating in the exam. Hence, any improvements or declines in achievements are relative to the mean and do not denote actual progress or deterioration over time. Nonetheless, it is still possible to gauge Israel s relative position visà-vis other countries and whether Egypt Botswana Jordan Lebanon South Africa Taiwan Saudi Arabia Iran Japan Qatar Australia UAE Slovenia Kazakhstan Norway United States Sweden Italy Singapore Israel Lithuania Georgia Hungary Malaysia Canada Bahrain Russia Malta Kuwait New Zealand England Korea Chile Hong Kong Turkey Ireland Oman Thailand Morocco 0% 5% Figure 17 OVERALL EXCLUSION RATES IN TIMSS 2015 SHARE OF PUPILS NOT PARTICIPATING IN MATHEMATICS EXAM 10% 15% 20% Data: TIMSS 2015 23% this position has improved or declined. Figure 18 provides composite achievements in the three main areas tested, math science, and reading, over the past several decades for 13 to 15-year olds on all international exams that Israel has participated in since 1963. On the face of it, Israel ranked number 1 http://shoresh.institute 22

in math in 1963 experienced a major decline in achievements ever since. However, it turns out that the 700 Figure 18 ISRAELI SCORES ON INTERNATIONAL EXAMS SECONDARY SCHOOLS (13-15 YEAR OLDS), 1963-2015 country excluded so many pupils and schools in 1963, that the results for that year are apparently without much basis in terms of reflecting the actual math level that existed in Israel then. Only since 1999 have Arab 650 600 550 500 450 math science reading Israeli children been included in the international exams on a regular basis. Israel s scores on the TIMSS 400 1963 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 Data: IEA, TIMSS and PISA exams have been consistently higher than in the PISA exams (Figure 19). This may be due to the fact that the share of developed countries participating in the TIMSS exams is much smaller than in the PISA exams, improving Israel s position relative to such a mean. Regardless of the exam however, there have been steady improvements relative to the means in the achievements of Israeli pupils whether in math, science or reading. Despite these improvements, average achievements in math, science and reading in the most recent PISA exam (administered in 2015) place Israeli children below the children of nearly all of the 25 520 500 480 460 440 Figure 19 ISRAELI SCORES ON PISA AND TIMSS EXAMS 1999-2015 math-timss science-timss reading-pisa math-pisa science-pisa relevant developed countries, with only Slovakian children attaining a 420 1999 2001 2003 2005 2007 2009 2011 2013 2015 PISA (15 year olds) and TIMSS (8th grade). lower score (Figure 20). One can Data: PISA and TIMSS http://shoresh.institute 23